Hardware security

[Plume]

As far as I understand it, it totally would be able to. It might be more or less efficient than a conventional computer (QC is not going to revolutionize the world of spreadsheets ...), but it shouldn't be impossible.
The problem with OTP is that if you intercept an encrypted text, trying out all combinations will produce a list of all texts with the same length.

 

[WizardStan]

That's called one-time pad and doesn't work with public-key crypto. If there is a public-key crypto that can't be (theoretically easily) broken by QC, I would like to know myself.

We had a researcher in to talk about post quantum cryptography a few weeks ago; most of it went entirely over my head, but he did talk about one public-key algorithm that was theoretically quantum safe in that it still scaled exponentially in quantum space, just not by as much as traditional computers. The end result would be a key that was substantially larger, and every time more q-bits were added it would get even larger still, but it was workable.

edit: I mention this to clarify that I have heard of such a thing that may or may not be viable based on my limited understanding of quantum computers, but I'll try to find the handout we were all given and let you have a look.

 

[Grench]

Quantum computers theoretically will solve all potential outcomes for all factors for a given key simultaneously.

Then it's a matter of picking which answer is the 'correct' one.

IMHO the next phase in encryption may actually be nested encryption with non-prime key generation using analog and/or in-the-world variable measurements as 2nd factors.

For example encrypting a file using an encryption/decryption engine that requires hardware level GPS access and will only decrypt if the device it is on is within 50 feet of where the public key was created. Yes, that could be spoofed - if you knew where the location of creation for the public key was. Since lat/long numbers are not required to be prime, the QC has lost a significant portion of it's advantage.

Another example: On a device like the Pyra, only allow decryption if a unique combination of keys and rough nub positioning that was used in creation of the public key is repeated.

Now layer in both of the above with a 2048bit encryption key. The generation of public keys with prime and non-prime factors (resulting in a larger multiples of prime factors) based on spot-measurement in key generation and decryption, where the solution requires finding a non-prime result with near analog generation. That'll keep 'em guessing for a bit.

 

[WizardStan]

Knowing a location, knowing some unique combination of keys, needing to share these with the target audience in order to decrypt; what you've described is effectively just a password, symmetric encryption. It's an entirely different class of security, and not one that can be solved by quantum computers so no one really cares. That's pretty useless for anonymous communication though: how can you securely transfer something to a website if you can't securely transfer the password, gps coordinates, or nub combination? That's why public key encryption is required.

 

[levi]

The end result would be a key that was substantially larger, and every time more q-bits were added it would get even larger still, but it was workable.

That fits with my understanding of pubkey crypto basics, and of how quantum computers work. I think you need enough qbits to represent the binary digits of both prime numbers so that you can multiply them together and get them to match something which represents the message.

But I don't really know how the prime numbers I'm talking about map on to the keys and message cyphers used by PGP/GPG. Actually, I think the two primes relate to your public and private keys, but now I don't understand how they can be used to send a message such that the receiver can't decode your private key, so I'm clearly still missing something. And I haven't got the foggiest how you actually program a quantum computer, so it deffo sounds like I need to do some more learning.

 

[Orbstheorem]

Basically, QC is able to solve most crypto problems that don't use a bijection from the origin set into the target set (or in general sets of the same size).

Can you point me to a scientific reference about this?
[doublepost=1464247053,1464246575][/doublepost]

That fits with my understanding of pubkey crypto basics, and of how quantum computers work. I think you need enough qbits to represent the binary digits of both prime numbers so that you can multiply them together and get them to match something which represents the message.

But I don't really know how the prime numbers I'm talking about map on to the keys and message cyphers used by PGP/GPG. Actually, I think the two primes relate to your public and private keys, but now I don't understand how they can be used to send a message such that the receiver can't decode your private key, so I'm clearly still missing something. And I haven't got the foggiest how you actually program a quantum computer, so it deffo sounds like I need to do some more learning.

Check this out, really helped on my Information Sciences Course: